*
**** Alterations to Henry's code are...
****
- **** Copyright (c) 1991-1997, Larry Wall
+ **** Copyright (c) 1991-1998, Larry Wall
****
**** You may distribute under the terms of either the GNU General Public
**** License or the Artistic License, as specified in the README file.
*/
#include "EXTERN.h"
#include "perl.h"
+typedef MAGIC *my_magic;
+
#include "regcomp.h"
#define RF_tainted 1 /* tainted information used? */
#define RF_warned 2 /* warned about big count? */
#define RF_evaled 4 /* Did an EVAL with setting? */
+#define RF_utf8 8 /* String contains multibyte chars? */
+
+#define UTF (PL_reg_flags & RF_utf8)
#define RS_init 1 /* eval environment created */
#define RS_set 2 /* replsv value is set */
static I32 regtry _((regexp *prog, char *startpos));
static bool reginclass _((char *p, I32 c));
+static bool reginclassutf8 _((regnode *f, U8* p));
static CHECKPOINT regcppush _((I32 parenfloor));
static char * regcppop _((void));
#endif
+
#define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c))
+#define REGINCLASSUTF8(f,p) (ARG1(f) ? reginclassutf8(f,p) : swash_fetch((SV*)PL_regdata->data[ARG2(f)],p))
+
+#define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv))
+#define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b)
+
+#ifndef PERL_OBJECT
+static U8 * reghop _((U8 *pos, I32 off));
+static U8 * reghopmaybe _((U8 *pos, I32 off));
+#endif
+#define reghop_c(pos,off) ((char*)reghop((U8*)pos, off))
+#define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off))
+#define HOP(pos,off) (UTF ? reghop((U8*)pos, off) : (U8*)(pos + off))
+#define HOPMAYBE(pos,off) (UTF ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off))
+#define HOPc(pos,off) ((char*)HOP(pos,off))
+#define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off))
STATIC CHECKPOINT
regcppush(I32 parenfloor)
return input;
}
+STATIC char *
+regcp_set_to(I32 ss)
+{
+ I32 tmp = PL_savestack_ix;
+
+ PL_savestack_ix = ss;
+ regcppop();
+ PL_savestack_ix = tmp;
+}
+
+typedef struct re_cc_state
+{
+ I32 ss;
+ regnode *node;
+ struct re_cc_state *prev;
+ CURCUR *cc;
+ regexp *re;
+} re_cc_state;
+
#define regcpblow(cp) LEAVE_SCOPE(cp)
/*
regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
nosave ? 0 : REXEC_COPY_STR);
}
+
+STATIC void
+cache_re(regexp *prog)
+{
+ PL_regprecomp = prog->precomp; /* Needed for FAIL. */
+#ifdef DEBUGGING
+ PL_regprogram = prog->program;
+#endif
+ PL_regnpar = prog->nparens;
+ PL_regdata = prog->data;
+ PL_reg_re = prog;
+}
/*
- regexec_flags - match a regexp against a string
I32 dontbother = 0; /* how many characters not to try at end */
CURCUR cc;
I32 start_shift = 0; /* Offset of the start to find
- constant substr. */
- I32 end_shift = 0; /* Same for the end. */
+ constant substr. */ /* CC */
+ I32 end_shift = 0; /* Same for the end. */ /* CC */
I32 scream_pos = -1; /* Internal iterator of scream. */
char *scream_olds;
SV* oreplsv = GvSV(PL_replgv);
cc.oldcc = 0;
PL_regcc = &cc;
- PL_regprecomp = prog->precomp; /* Needed for error messages. */
+ cache_re(prog);
#ifdef DEBUGGING
PL_regnarrate = PL_debug & 512;
- PL_regprogram = prog->program;
#endif
/* Be paranoid... */
if (startpos == strbeg) /* is ^ valid at stringarg? */
PL_regprev = '\n';
else {
- PL_regprev = stringarg[-1];
+ PL_regprev = (U32)stringarg[-1];
if (!PL_multiline && PL_regprev == '\n')
PL_regprev = '\0'; /* force ^ to NOT match */
}
FAIL("corrupted regexp program");
}
- PL_regnpar = prog->nparens;
PL_reg_flags = 0;
PL_reg_eval_set = 0;
+ if (prog->reganch & ROPT_UTF8)
+ PL_reg_flags |= RF_utf8;
+
+ /* Mark beginning of line for ^ and lookbehind. */
+ PL_regbol = startpos;
+ PL_bostr = strbeg;
+
+ /* Mark end of line for $ (and such) */
+ PL_regeol = strend;
+
+ /* see how far we have to get to not match where we matched before */
+ PL_regtill = startpos+minend;
+
+ /* We start without call_cc context. */
+ PL_reg_call_cc = 0;
+
/* If there is a "must appear" string, look for it. */
s = startpos;
if (!(flags & REXEC_CHECKED)
(!(prog->reganch & (ROPT_ANCH_BOL | ROPT_ANCH_MBOL))
|| (PL_multiline && prog->check_substr == prog->anchored_substr)) )
{
- start_shift = prog->check_offset_min;
+ char *t;
+ start_shift = prog->check_offset_min; /* okay to underestimate on CC */
/* Should be nonnegative! */
- end_shift = minlen - start_shift - SvCUR(prog->check_substr);
+ end_shift = minlen - start_shift - CHR_SVLEN(prog->check_substr);
if (screamer) {
if (PL_screamfirst[BmRARE(prog->check_substr)] >= 0)
s = screaminstr(screamer, prog->check_substr,
if (!s) {
++BmUSEFUL(prog->check_substr); /* hooray */
goto phooey; /* not present */
- } else if ((s - stringarg) > prog->check_offset_max) {
+ }
+ else if (s - stringarg > prog->check_offset_max &&
+ (UTF
+ ? ((t = reghopmaybe_c(s, -(prog->check_offset_max))) && t >= stringarg)
+ : (t = s - prog->check_offset_max) != 0
+ )
+ )
+ {
++BmUSEFUL(prog->check_substr); /* hooray/2 */
- s -= prog->check_offset_max;
- } else if (!prog->naughty
+ s = t;
+ }
+ else if (!(prog->reganch & ROPT_NAUGHTY)
&& --BmUSEFUL(prog->check_substr) < 0
&& prog->check_substr == prog->float_substr) { /* boo */
SvREFCNT_dec(prog->check_substr);
prog->check_substr = Nullsv; /* disable */
prog->float_substr = Nullsv; /* clear */
s = startpos;
- } else s = startpos;
+ }
+ else
+ s = startpos;
}
- /* Mark beginning of line for ^ and lookbehind. */
- PL_regbol = startpos;
- PL_bostr = strbeg;
-
- /* Mark end of line for $ (and such) */
- PL_regeol = strend;
-
- /* see how far we have to get to not match where we matched before */
- PL_regtill = startpos+minend;
-
DEBUG_r(
PerlIO_printf(Perl_debug_log,
- "Matching `%.60s%s' against `%.*s%s'\n",
- prog->precomp,
+ "%sMatching%s `%s%.60s%s%s' against `%s%.*s%s%s'\n",
+ PL_colors[4],PL_colors[5],PL_colors[0],
+ prog->precomp,
+ PL_colors[1],
(strlen(prog->precomp) > 60 ? "..." : ""),
+ PL_colors[0],
(strend - startpos > 60 ? 60 : strend - startpos),
- startpos,
+ startpos, PL_colors[1],
(strend - startpos > 60 ? "..." : ""))
);
{
if (minlen)
dontbother = minlen - 1;
- strend -= dontbother;
+ strend = HOPc(strend, -dontbother);
/* for multiline we only have to try after newlines */
if (s > startpos)
s--;
while (s < strend) {
- if (*s++ == '\n') {
+ if (*s++ == '\n') { /* don't need utf8skip here */
if (s < strend && regtry(prog, s))
goto got_it;
}
/* we have /x+whatever/ */
/* it must be a one character string */
char ch = SvPVX(prog->anchored_substr)[0];
- while (s < strend) {
- if (*s == ch) {
- if (regtry(prog, s)) goto got_it;
- s++;
- while (s < strend && *s == ch)
+ if (UTF) {
+ while (s < strend) {
+ if (*s == ch) {
+ if (regtry(prog, s)) goto got_it;
+ s += UTF8SKIP(s);
+ while (s < strend && *s == ch)
+ s += UTF8SKIP(s);
+ }
+ s += UTF8SKIP(s);
+ }
+ }
+ else {
+ while (s < strend) {
+ if (*s == ch) {
+ if (regtry(prog, s)) goto got_it;
s++;
+ while (s < strend && *s == ch)
+ s++;
+ }
+ s++;
}
- s++;
}
}
/*SUPPRESS 560*/
I32 back_min =
prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset;
I32 delta = back_max - back_min;
- char *last = strend - SvCUR(must) - back_min; /* Cannot start after this */
- char *last1 = s - 1; /* Last position checked before */
+ char *last = HOPc(strend, 0-(CHR_SVLEN(must) + back_min)); /* Cannot start after this */
+ char *last1; /* Last position checked before */
+
+ if (s > PL_bostr)
+ last1 = HOPc(s, -1);
+ else
+ last1 = s - 1; /* bogus */
/* XXXX check_substr already used to find `s', can optimize if
check_substr==must. */
scream_pos = -1;
dontbother = end_shift;
- strend -= dontbother;
+ strend = HOPc(strend, -dontbother);
while ( (s <= last) &&
(screamer
- ? (s = screaminstr(screamer, must, s + back_min - strbeg,
+ ? (s = screaminstr(screamer, must, HOPc(s, back_min) - strbeg,
end_shift, &scream_pos, 0))
- : (s = fbm_instr((unsigned char*)s + back_min,
+ : (s = fbm_instr((unsigned char*)HOP(s, back_min),
(unsigned char*)strend, must, 0))) ) {
- if (s - back_max > last1) {
- last1 = s - back_min;
- s = s - back_max;
- } else {
- char *t = last1 + 1;
+ if (HOPc(s, -back_max) > last1) {
+ last1 = HOPc(s, -back_min);
+ s = HOPc(s, -back_max);
+ }
+ else {
+ char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1;
- last1 = s - back_min;
+ last1 = HOPc(s, -back_min);
s = t;
}
- while (s <= last1) {
- if (regtry(prog, s))
- goto got_it;
- s++;
+ if (UTF) {
+ while (s <= last1) {
+ if (regtry(prog, s))
+ goto got_it;
+ s += UTF8SKIP(s);
+ }
+ }
+ else {
+ while (s <= last1) {
+ if (regtry(prog, s))
+ goto got_it;
+ s++;
+ }
}
}
goto phooey;
- } else if (c = prog->regstclass) {
+ }
+ else if (c = prog->regstclass) {
I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
- char *Class;
+ char *cc;
if (minlen)
dontbother = minlen - 1;
- strend -= dontbother; /* don't bother with what can't match */
+ strend = HOPc(strend, -dontbother); /* don't bother with what can't match */
tmp = 1;
/* We know what class it must start with. */
switch (OP(c)) {
+ case ANYOFUTF8:
+ cc = (char *) OPERAND(c);
+ while (s < strend) {
+ if (REGINCLASSUTF8(c, (U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case ANYOF:
- Class = (char *) OPERAND(c);
+ cc = (char *) OPERAND(c);
while (s < strend) {
- if (REGINCLASS(Class, *s)) {
+ if (REGINCLASS(cc, *s)) {
if (tmp && regtry(prog, s))
goto got_it;
else
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case BOUND:
- if (minlen)
- dontbother++,strend--;
+ if (minlen) {
+ dontbother++;
+ strend -= 1;
+ }
tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev;
tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
while (s < strend) {
if ((minlen || tmp) && regtry(prog,s))
goto got_it;
break;
+ case BOUNDLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case BOUNDUTF8:
+ if (minlen) {
+ dontbother++;
+ strend = reghop_c(strend, -1);
+ }
+ tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev;
+ tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
+ while (s < strend) {
+ if (tmp == !(OP(c) == BOUND ?
+ swash_fetch(PL_utf8_alnum, (U8*)s) :
+ isALNUM_LC_utf8((U8*)s)))
+ {
+ tmp = !tmp;
+ if (regtry(prog, s))
+ goto got_it;
+ }
+ s += UTF8SKIP(s);
+ }
+ if ((minlen || tmp) && regtry(prog,s))
+ goto got_it;
+ break;
case NBOUNDL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
case NBOUND:
- if (minlen)
- dontbother++,strend--;
+ if (minlen) {
+ dontbother++;
+ strend -= 1;
+ }
tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev;
tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0);
while (s < strend) {
if ((minlen || !tmp) && regtry(prog,s))
goto got_it;
break;
+ case NBOUNDLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NBOUNDUTF8:
+ if (minlen) {
+ dontbother++;
+ strend = reghop_c(strend, -1);
+ }
+ tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev;
+ tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0);
+ while (s < strend) {
+ if (tmp == !(OP(c) == NBOUND ?
+ swash_fetch(PL_utf8_alnum, (U8*)s) :
+ isALNUM_LC_utf8((U8*)s)))
+ tmp = !tmp;
+ else if (regtry(prog, s))
+ goto got_it;
+ s += UTF8SKIP(s);
+ }
+ if ((minlen || !tmp) && regtry(prog,s))
+ goto got_it;
+ break;
case ALNUM:
while (s < strend) {
if (isALNUM(*s)) {
s++;
}
break;
+ case ALNUMUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_alnum, (U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case ALNUML:
PL_reg_flags |= RF_tainted;
while (s < strend) {
s++;
}
break;
+ case ALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (isALNUM_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case NALNUM:
while (s < strend) {
if (!isALNUM(*s)) {
s++;
}
break;
+ case NALNUMUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_alnum, (U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case NALNUML:
PL_reg_flags |= RF_tainted;
while (s < strend) {
s++;
}
break;
+ case NALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!isALNUM_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case SPACE:
while (s < strend) {
if (isSPACE(*s)) {
s++;
}
break;
+ case SPACEUTF8:
+ while (s < strend) {
+ if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case SPACEL:
PL_reg_flags |= RF_tainted;
while (s < strend) {
s++;
}
break;
+ case SPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case NSPACE:
while (s < strend) {
if (!isSPACE(*s)) {
s++;
}
break;
+ case NSPACEUTF8:
+ while (s < strend) {
+ if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case NSPACEL:
PL_reg_flags |= RF_tainted;
while (s < strend) {
s++;
}
break;
+ case NSPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ while (s < strend) {
+ if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case DIGIT:
while (s < strend) {
if (isDIGIT(*s)) {
s++;
}
break;
+ case DIGITUTF8:
+ while (s < strend) {
+ if (swash_fetch(PL_utf8_digit,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
case NDIGIT:
while (s < strend) {
if (!isDIGIT(*s)) {
s++;
}
break;
+ case NDIGITUTF8:
+ while (s < strend) {
+ if (!swash_fetch(PL_utf8_digit,(U8*)s)) {
+ if (tmp && regtry(prog, s))
+ goto got_it;
+ else
+ tmp = doevery;
+ }
+ else
+ tmp = 1;
+ s += UTF8SKIP(s);
+ }
+ break;
}
}
else {
if (!last) {
last = scream_olds; /* Only one occurence. */
}
- } else {
+ }
+ else {
STRLEN len;
char *little = SvPV(prog->float_substr, len);
if (len)
}
if (minlen && (dontbother < minlen))
dontbother = minlen - 1;
- strend -= dontbother;
+ strend -= dontbother; /* this one's always in bytes! */
/* We don't know much -- general case. */
- do {
- if (regtry(prog, s))
- goto got_it;
- } while (s++ < strend);
+ if (UTF) {
+ for (;;) {
+ if (regtry(prog, s))
+ goto got_it;
+ if (s >= strend)
+ break;
+ s += UTF8SKIP(s);
+ };
+ }
+ else {
+ do {
+ if (regtry(prog, s))
+ goto got_it;
+ } while (s++ < strend);
+ }
}
/* Failure. */
sp = prog->startp;
ep = prog->endp;
- PL_regdata = prog->data;
if (prog->nparens) {
for (i = prog->nparens; i >= 0; i--) {
*sp++ = NULL;
PL_regindent++;
#endif
+ /* Note that nextchr is a byte even in UTF */
nextchr = UCHARAT(locinput);
scan = prog;
while (scan != NULL) {
int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput);
int pref_len = (locinput - PL_bostr > (5 + taill) - l
? (5 + taill) - l : locinput - PL_bostr);
+ int pref0_len = pref_len - (locinput - PL_reginput);
if (l + pref_len < (5 + taill) && l < PL_regeol - locinput)
l = ( PL_regeol - locinput > (5 + taill) - pref_len
? (5 + taill) - pref_len : PL_regeol - locinput);
+ if (pref0_len < 0)
+ pref0_len = 0;
regprop(prop, scan);
PerlIO_printf(Perl_debug_log,
- "%4i <%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n",
+ "%4i <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n",
locinput - PL_bostr,
- PL_colors[2], pref_len, locinput - pref_len, PL_colors[3],
+ PL_colors[4], pref0_len,
+ locinput - pref_len, PL_colors[5],
+ PL_colors[2], pref_len - pref0_len,
+ locinput - pref_len + pref0_len, PL_colors[3],
(docolor ? "" : "> <"),
PL_colors[0], l, locinput, PL_colors[1],
15 - l - pref_len + 1,
: (PL_multiline &&
(nextchr || locinput < PL_regeol) && locinput[-1] == '\n') )
{
- /* PL_regtill = PL_regbol; */
+ /* regtill = regbol; */
break;
}
sayNO;
if (PL_regeol != locinput)
sayNO;
break;
+ case SANYUTF8:
+ if (nextchr & 0x80) {
+ locinput += utf8skip[nextchr];
+ if (locinput > PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
case SANY:
if (!nextchr && locinput >= PL_regeol)
sayNO;
nextchr = UCHARAT(++locinput);
break;
+ case ANYUTF8:
+ if (nextchr & 0x80) {
+ locinput += utf8skip[nextchr];
+ if (locinput > PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
case ANY:
if (!nextchr && locinput >= PL_regeol || nextchr == '\n')
sayNO;
case EXACTF:
s = (char *) OPERAND(scan);
ln = UCHARAT(s++);
+
+ if (UTF) {
+ char *l = locinput;
+ char *e = s + ln;
+ c1 = OP(scan) == EXACTF;
+ while (s < e) {
+ if (l >= PL_regeol)
+ sayNO;
+ if (utf8_to_uv((U8*)s, 0) != (c1 ?
+ toLOWER_utf8((U8*)l) :
+ toLOWER_LC_utf8((U8*)l)))
+ {
+ sayNO;
+ }
+ s += UTF8SKIP(s);
+ l += UTF8SKIP(l);
+ }
+ locinput = l;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+
/* Inline the first character, for speed. */
if (UCHARAT(s) != nextchr &&
UCHARAT(s) != ((OP(scan) == EXACTF)
? ibcmp(s, locinput, ln)
: ibcmp_locale(s, locinput, ln)))
sayNO;
- locinput += ln;
+ locinput += ln;
+ nextchr = UCHARAT(locinput);
+ break;
+ case ANYOFUTF8:
+ s = (char *) OPERAND(scan);
+ if (!REGINCLASSUTF8(scan, (U8*)locinput))
+ sayNO;
+ if (locinput >= PL_regeol)
+ sayNO;
+ locinput += utf8skip[nextchr];
nextchr = UCHARAT(locinput);
break;
case ANYOF:
sayNO;
nextchr = UCHARAT(++locinput);
break;
+ case ALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case ALNUMUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == ALNUMUTF8
+ ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
+ : isALNUM_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == ALNUMUTF8
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
case NALNUML:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
sayNO;
nextchr = UCHARAT(++locinput);
break;
+ case NALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NALNUMUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (OP(scan) == NALNUMUTF8
+ ? swash_fetch(PL_utf8_alnum, (U8*)locinput)
+ : isALNUM_LC_utf8((U8*)locinput))
+ {
+ sayNO;
+ }
+ locinput += utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (OP(scan) == NALNUMUTF8
+ ? isALNUM(nextchr) : isALNUM_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
case BOUNDL:
case NBOUNDL:
PL_reg_flags |= RF_tainted;
if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL))
sayNO;
break;
+ case BOUNDLUTF8:
+ case NBOUNDLUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case BOUNDUTF8:
+ case NBOUNDUTF8:
+ /* was last char in word? */
+ ln = (locinput != PL_regbol)
+ ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev;
+ if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) {
+ ln = isALNUM_uni(ln);
+ n = swash_fetch(PL_utf8_alnum, (U8*)locinput);
+ }
+ else {
+ ln = isALNUM_LC_uni(ln);
+ n = isALNUM_LC_utf8((U8*)locinput);
+ }
+ if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8))
+ sayNO;
+ break;
case SPACEL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
sayNO;
nextchr = UCHARAT(++locinput);
break;
+ case SPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case SPACEUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (!(OP(scan) == SPACEUTF8
+ ? swash_fetch(PL_utf8_space,(U8*)locinput)
+ : isSPACE_LC_utf8((U8*)locinput)))
+ {
+ sayNO;
+ }
+ locinput += utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!(OP(scan) == SPACEUTF8
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr)))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
case NSPACEL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
sayNO;
nextchr = UCHARAT(++locinput);
break;
+ case NSPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ /* FALL THROUGH */
+ case NSPACEUTF8:
+ if (!nextchr)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (OP(scan) == NSPACEUTF8
+ ? swash_fetch(PL_utf8_space,(U8*)locinput)
+ : isSPACE_LC_utf8((U8*)locinput))
+ {
+ sayNO;
+ }
+ locinput += utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (OP(scan) == NSPACEUTF8
+ ? isSPACE(nextchr) : isSPACE_LC(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
case DIGIT:
if (!isDIGIT(nextchr))
sayNO;
nextchr = UCHARAT(++locinput);
break;
+ case DIGITUTF8:
+ if (nextchr & 0x80) {
+ if (!(swash_fetch(PL_utf8_digit,(U8*)locinput)))
+ sayNO;
+ locinput += utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (!isDIGIT(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
case NDIGIT:
if (!nextchr && locinput >= PL_regeol)
sayNO;
sayNO;
nextchr = UCHARAT(++locinput);
break;
+ case NDIGITUTF8:
+ if (!nextchr && locinput >= PL_regeol)
+ sayNO;
+ if (nextchr & 0x80) {
+ if (swash_fetch(PL_utf8_digit,(U8*)locinput))
+ sayNO;
+ locinput += utf8skip[nextchr];
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+ if (isDIGIT(nextchr))
+ sayNO;
+ nextchr = UCHARAT(++locinput);
+ break;
+ case CLUMP:
+ if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput))
+ sayNO;
+ locinput += utf8skip[nextchr];
+ while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput))
+ locinput += UTF8SKIP(locinput);
+ if (locinput > PL_regeol)
+ sayNO;
+ nextchr = UCHARAT(locinput);
+ break;
case REFFL:
PL_reg_flags |= RF_tainted;
/* FALL THROUGH */
sayNO; /* Do not match unless seen CLOSEn. */
if (s == PL_regendp[n])
break;
+
+ if (UTF && OP(scan) != REF) { /* REF can do byte comparison */
+ char *l = locinput;
+ char *e = PL_regendp[n];
+ /*
+ * Note that we can't do the "other character" lookup trick as
+ * in the 8-bit case (no pun intended) because in Unicode we
+ * have to map both upper and title case to lower case.
+ */
+ if (OP(scan) == REFF) {
+ while (s < e) {
+ if (l >= PL_regeol)
+ sayNO;
+ if (toLOWER_utf8((U8*)s) != toLOWER_utf8((U8*)l))
+ sayNO;
+ s += UTF8SKIP(s);
+ l += UTF8SKIP(l);
+ }
+ }
+ else {
+ while (s < e) {
+ if (l >= PL_regeol)
+ sayNO;
+ if (toLOWER_LC_utf8((U8*)s) != toLOWER_LC_utf8((U8*)l))
+ sayNO;
+ s += UTF8SKIP(s);
+ l += UTF8SKIP(l);
+ }
+ }
+ locinput = l;
+ nextchr = UCHARAT(locinput);
+ break;
+ }
+
/* Inline the first character, for speed. */
if (UCHARAT(s) != nextchr &&
(OP(scan) == REF ||
ret = POPs;
PUTBACK;
- if (logical) {
- logical = 0;
- sw = SvTRUE(ret);
- } else
- sv_setsv(save_scalar(PL_replgv), ret);
PL_op = oop;
PL_curpad = ocurpad;
PL_curcop = ocurcop;
+ if (logical) {
+ if (logical == 2) { /* Postponed subexpression. */
+ regexp *re;
+ my_magic mg = Null(my_magic);
+ re_cc_state state;
+ CURCUR cctmp;
+ CHECKPOINT cp, lastcp;
+
+ if(SvROK(ret) || SvRMAGICAL(ret)) {
+ SV *sv = SvROK(ret) ? SvRV(ret) : ret;
+
+ if(SvMAGICAL(sv))
+ mg = mg_find(sv, 'r');
+ }
+ if (mg) {
+ re = (regexp *)mg->mg_obj;
+ ReREFCNT_inc(re);
+ }
+ else {
+ STRLEN len;
+ char *t = SvPV(ret, len);
+ PMOP pm;
+ char *oprecomp = PL_regprecomp;
+ I32 osize = PL_regsize;
+ I32 onpar = PL_regnpar;
+
+ pm.op_pmflags = 0;
+ re = CALLREGCOMP(t, t + len, &pm);
+ if (!(SvFLAGS(ret)
+ & (SVs_TEMP | SVs_PADTMP | SVf_READONLY)))
+ sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0);
+ PL_regprecomp = oprecomp;
+ PL_regsize = osize;
+ PL_regnpar = onpar;
+ }
+ DEBUG_r(
+ PerlIO_printf(Perl_debug_log,
+ "Entering embedded `%s%.60s%s%s'\n",
+ PL_colors[0],
+ re->precomp,
+ PL_colors[1],
+ (strlen(re->precomp) > 60 ? "..." : ""))
+ );
+ state.node = next;
+ state.prev = PL_reg_call_cc;
+ state.cc = PL_regcc;
+ state.re = PL_reg_re;
+
+ cctmp.cur = 0;
+ cctmp.oldcc = 0;
+ PL_regcc = &cctmp;
+
+ cp = regcppush(0); /* Save *all* the positions. */
+ REGCP_SET;
+ cache_re(re);
+ state.ss = PL_savestack_ix;
+ *PL_reglastparen = 0;
+ PL_reg_call_cc = &state;
+ PL_reginput = locinput;
+ if (regmatch(re->program + 1)) {
+ ReREFCNT_dec(re);
+ regcpblow(cp);
+ sayYES;
+ }
+ DEBUG_r(
+ PerlIO_printf(Perl_debug_log,
+ "%*s failed...\n",
+ REPORT_CODE_OFF+PL_regindent*2, "")
+ );
+ ReREFCNT_dec(re);
+ REGCP_UNWIND;
+ regcppop();
+ PL_reg_call_cc = state.prev;
+ PL_regcc = state.cc;
+ PL_reg_re = state.re;
+ sayNO;
+ }
+ sw = SvTRUE(ret);
+ logical = 0;
+ }
+ else
+ sv_setsv(save_scalar(PL_replgv), ret);
break;
}
case OPEN:
}
break;
case LOGICAL:
- logical = 1;
+ logical = scan->flags;
break;
case CURLYX: {
CURCUR cc;
PL_regcc = cc;
if (n >= cc->max) { /* Maximum greed exceeded? */
- if (PL_dowarn && n >= REG_INFTY
+ if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
&& !(PL_reg_flags & RF_warned)) {
PL_reg_flags |= RF_warned;
- warn("%s limit (%d) exceeded",
+ warner(WARN_UNSAFE, "%s limit (%d) exceeded",
"Complex regular subexpression recursion",
REG_INFTY - 1);
}
REPORT_CODE_OFF+PL_regindent*2, "")
);
}
- if (PL_dowarn && n >= REG_INFTY && !(PL_reg_flags & RF_warned)) {
+ if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY
+ && !(PL_reg_flags & RF_warned)) {
PL_reg_flags |= RF_warned;
- warn("count exceeded %d", REG_INFTY - 1);
+ warner(WARN_UNSAFE, "%s limit (%d) exceeded",
+ "Complex regular subexpression recursion",
+ REG_INFTY - 1);
}
/* Failed deeper matches of scan, so see if this one works. */
c2 = fold_locale[c1];
else
c2 = c1;
- } else
+ }
+ else
c1 = c2 = -1000;
REGCP_SET;
/* This may be improved if l == 0. */
{
if (paren) {
if (n) {
- PL_regstartp[paren] = PL_reginput - l;
+ PL_regstartp[paren] = HOPc(PL_reginput, -l);
PL_regendp[paren] = PL_reginput;
- } else
+ }
+ else
PL_regendp[paren] = NULL;
}
if (regmatch(next))
else
sayNO;
}
- } else {
+ }
+ else {
n = regrepeat_hard(scan, n, &l);
if (n != 0 && l == 0
/* In fact, this is tricky. If paren, then the
c2 = fold_locale[c1];
else
c2 = c1;
- } else
+ }
+ else
c1 = c2 = -1000;
}
REGCP_SET;
if (c1 == -1000 ||
UCHARAT(PL_reginput) == c1 ||
UCHARAT(PL_reginput) == c2)
- {
- DEBUG_r(
+ {
+ DEBUG_r(
PerlIO_printf(Perl_debug_log,
"%*s trying tail with n=%ld...\n",
REPORT_CODE_OFF+PL_regindent*2, "", n)
- );
- if (paren) {
- if (n) {
- PL_regstartp[paren] = PL_reginput - l;
- PL_regendp[paren] = PL_reginput;
- } else
- PL_regendp[paren] = NULL;
+ );
+ if (paren) {
+ if (n) {
+ PL_regstartp[paren] = HOPc(PL_reginput, -l);
+ PL_regendp[paren] = PL_reginput;
}
- if (regmatch(next))
- sayYES;
- REGCP_UNWIND;
+ else
+ PL_regendp[paren] = NULL;
}
+ if (regmatch(next))
+ sayYES;
+ REGCP_UNWIND;
+ }
/* Couldn't or didn't -- back up. */
n--;
- locinput -= l;
+ locinput = HOPc(locinput, -l);
PL_reginput = locinput;
}
}
minmod = 0;
if (ln && regrepeat(scan, ln) < ln)
sayNO;
+ locinput = PL_reginput;
REGCP_SET;
while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */
/* If it could work, try it. */
{
if (paren) {
if (n) {
- PL_regstartp[paren] = PL_reginput - 1;
+ PL_regstartp[paren] = HOPc(PL_reginput, -1);
PL_regendp[paren] = PL_reginput;
- } else
+ }
+ else
PL_regendp[paren] = NULL;
}
if (regmatch(next))
REGCP_UNWIND;
}
/* Couldn't or didn't -- move forward. */
- PL_reginput = locinput + ln;
+ PL_reginput = locinput;
if (regrepeat(scan, 1)) {
ln++;
- PL_reginput = locinput + ln;
- } else
+ locinput = PL_reginput;
+ }
+ else
sayNO;
}
}
else {
CHECKPOINT lastcp;
n = regrepeat(scan, n);
+ locinput = PL_reginput;
if (ln < n && regkind[(U8)OP(next)] == EOL &&
(!PL_multiline || OP(next) == SEOL))
ln = n; /* why back off? */
{
if (paren && n) {
if (n) {
- PL_regstartp[paren] = PL_reginput - 1;
+ PL_regstartp[paren] = HOPc(PL_reginput, -1);
PL_regendp[paren] = PL_reginput;
- } else
+ }
+ else
PL_regendp[paren] = NULL;
}
if (regmatch(next))
}
/* Couldn't or didn't -- back up. */
n--;
- PL_reginput = locinput + n;
+ PL_reginput = locinput = HOPc(locinput, -1);
}
- } else {
+ }
+ else {
while (n >= ln) {
/* If it could work, try it. */
if (c1 == -1000 ||
}
/* Couldn't or didn't -- back up. */
n--;
- PL_reginput = locinput + n;
+ PL_reginput = locinput = HOPc(locinput, -1);
}
}
}
sayNO;
break;
case END:
+ if (PL_reg_call_cc) {
+ re_cc_state *cur_call_cc = PL_reg_call_cc;
+ CURCUR *cctmp = PL_regcc;
+ regexp *re = PL_reg_re;
+ CHECKPOINT cp, lastcp;
+
+ cp = regcppush(0); /* Save *all* the positions. */
+ REGCP_SET;
+ regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of
+ the caller. */
+ PL_reginput = locinput; /* Make position available to
+ the callcc. */
+ cache_re(PL_reg_call_cc->re);
+ PL_regcc = PL_reg_call_cc->cc;
+ PL_reg_call_cc = PL_reg_call_cc->prev;
+ if (regmatch(cur_call_cc->node)) {
+ PL_reg_call_cc = cur_call_cc;
+ regcpblow(cp);
+ sayYES;
+ }
+ REGCP_UNWIND;
+ regcppop();
+ PL_reg_call_cc = cur_call_cc;
+ PL_regcc = cctmp;
+ PL_reg_re = re;
+ cache_re(re);
+
+ DEBUG_r(
+ PerlIO_printf(Perl_debug_log,
+ "%*s continuation failed...\n",
+ REPORT_CODE_OFF+PL_regindent*2, "")
+ );
+ sayNO;
+ }
if (locinput < PL_regtill)
sayNO; /* Cannot match: too short. */
/* Fall through */
sayYES; /* Success! */
case SUSPEND:
n = 1;
+ PL_reginput = locinput;
goto do_ifmatch;
case UNLESSM:
n = 0;
- if (locinput < PL_bostr + scan->flags)
- goto say_yes;
+ if (scan->flags) {
+ s = HOPMAYBEc(locinput, -scan->flags);
+ if (!s)
+ goto say_yes;
+ PL_reginput = s;
+ }
+ else
+ PL_reginput = locinput;
goto do_ifmatch;
case IFMATCH:
n = 1;
- if (locinput < PL_bostr + scan->flags)
- goto say_no;
+ if (scan->flags) {
+ s = HOPMAYBEc(locinput, -scan->flags);
+ if (!s)
+ goto say_no;
+ PL_reginput = s;
+ }
+ else
+ PL_reginput = locinput;
+
do_ifmatch:
- PL_reginput = locinput - scan->flags;
inner = NEXTOPER(NEXTOPER(scan));
if (regmatch(inner) != n) {
say_no:
logical = 0;
sw = 0;
goto do_longjump;
- } else
+ }
+ else
sayNO;
}
say_yes:
/*
* [This routine now assumes that it will only match on things of length 1.
* That was true before, but now we assume scan - reginput is the count,
- * rather than incrementing count on every character.]
+ * rather than incrementing count on every character. [Er, except utf8.]]
*/
STATIC I32
regrepeat(regnode *p, I32 max)
register char *opnd;
register I32 c;
register char *loceol = PL_regeol;
+ register I32 hardcount = 0;
scan = PL_reginput;
if (max != REG_INFTY && max < loceol - scan)
case SANY:
scan = loceol;
break;
+ case ANYUTF8:
+ loceol = PL_regeol;
+ while (scan < loceol && *scan != '\n') {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
+ case SANYUTF8:
+ loceol = PL_regeol;
+ while (scan < loceol) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
case EXACT: /* length of string is 1 */
c = UCHARAT(++opnd);
while (scan < loceol && UCHARAT(scan) == c)
(UCHARAT(scan) == c || UCHARAT(scan) == fold_locale[c]))
scan++;
break;
+ case ANYOFUTF8:
+ loceol = PL_regeol;
+ while (scan < loceol && REGINCLASSUTF8(p, (U8*)scan)) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
case ANYOF:
while (scan < loceol && REGINCLASS(opnd, *scan))
scan++;
while (scan < loceol && isALNUM(*scan))
scan++;
break;
+ case ALNUMUTF8:
+ loceol = PL_regeol;
+ while (scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
case ALNUML:
PL_reg_flags |= RF_tainted;
while (scan < loceol && isALNUM_LC(*scan))
scan++;
break;
+ case ALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ loceol = PL_regeol;
+ while (scan < loceol && isALNUM_LC_utf8((U8*)scan)) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
+ break;
case NALNUM:
while (scan < loceol && !isALNUM(*scan))
scan++;
break;
+ case NALNUMUTF8:
+ loceol = PL_regeol;
+ while (scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
case NALNUML:
PL_reg_flags |= RF_tainted;
while (scan < loceol && !isALNUM_LC(*scan))
scan++;
break;
+ case NALNUMLUTF8:
+ PL_reg_flags |= RF_tainted;
+ loceol = PL_regeol;
+ while (scan < loceol && !isALNUM_LC_utf8((U8*)scan)) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
case SPACE:
while (scan < loceol && isSPACE(*scan))
scan++;
break;
+ case SPACEUTF8:
+ loceol = PL_regeol;
+ while (scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
case SPACEL:
PL_reg_flags |= RF_tainted;
while (scan < loceol && isSPACE_LC(*scan))
scan++;
break;
+ case SPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ loceol = PL_regeol;
+ while (scan < loceol && (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
case NSPACE:
while (scan < loceol && !isSPACE(*scan))
scan++;
break;
+ case NSPACEUTF8:
+ loceol = PL_regeol;
+ while (scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
case NSPACEL:
PL_reg_flags |= RF_tainted;
while (scan < loceol && !isSPACE_LC(*scan))
scan++;
break;
+ case NSPACELUTF8:
+ PL_reg_flags |= RF_tainted;
+ loceol = PL_regeol;
+ while (scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
case DIGIT:
while (scan < loceol && isDIGIT(*scan))
scan++;
break;
+ case DIGITUTF8:
+ loceol = PL_regeol;
+ while (scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
+ break;
case NDIGIT:
while (scan < loceol && !isDIGIT(*scan))
scan++;
break;
+ case NDIGITUTF8:
+ loceol = PL_regeol;
+ while (scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) {
+ scan += UTF8SKIP(scan);
+ hardcount++;
+ }
+ break;
default: /* Called on something of 0 width. */
break; /* So match right here or not at all. */
}
- c = scan - PL_reginput;
+ if (hardcount)
+ c = hardcount;
+ else
+ c = scan - PL_reginput;
PL_reginput = scan;
DEBUG_r(
register char *scan;
register char *start;
register char *loceol = PL_regeol;
- I32 l = -1;
+ I32 l = 0;
+ I32 count = 0, res = 1;
+
+ if (!max)
+ return 0;
start = PL_reginput;
- while (PL_reginput < loceol && (scan = PL_reginput, regmatch(p))) {
- if (l == -1) {
- *lp = l = PL_reginput - start;
- if (max != REG_INFTY && l*max < loceol - scan)
- loceol = scan + l*max;
- if (l == 0) {
- return max;
+ if (UTF) {
+ while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) {
+ if (!count++) {
+ l = 0;
+ while (start < PL_reginput) {
+ l++;
+ start += UTF8SKIP(start);
+ }
+ *lp = l;
+ if (l == 0)
+ return max;
}
+ if (count == max)
+ return count;
}
}
- if (PL_reginput < loceol)
+ else {
+ while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) {
+ if (!count++) {
+ *lp = l = PL_reginput - start;
+ if (max != REG_INFTY && l*max < loceol - scan)
+ loceol = scan + l*max;
+ if (l == 0)
+ return max;
+ }
+ }
+ }
+ if (!res)
PL_reginput = scan;
- else
- scan = PL_reginput;
- return (scan - start)/l;
+ return count;
}
/*
return (flags & ANYOF_INVERT) ? !match : match;
}
+STATIC bool
+reginclassutf8(regnode *f, U8 *p)
+{
+ dTHR;
+ char flags = ARG1(f);
+ bool match = FALSE;
+ SV *sv = (SV*)PL_regdata->data[ARG2(f)];
+
+ if (swash_fetch(sv, p))
+ match = TRUE;
+ else if (flags & ANYOF_FOLD) {
+ I32 cf;
+ U8 tmpbuf[10];
+ if (flags & ANYOF_LOCALE) {
+ PL_reg_flags |= RF_tainted;
+ uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p));
+ }
+ else
+ uv_to_utf8(tmpbuf, toLOWER_utf8(p));
+ if (swash_fetch(sv, tmpbuf))
+ match = TRUE;
+ }
+
+ if (!match && (flags & ANYOF_ISA)) {
+ PL_reg_flags |= RF_tainted;
+
+ if (((flags & ANYOF_ALNUML) && isALNUM_LC_utf8(p)) ||
+ ((flags & ANYOF_NALNUML) && !isALNUM_LC_utf8(p)) ||
+ ((flags & ANYOF_SPACEL) && isSPACE_LC_utf8(p)) ||
+ ((flags & ANYOF_NSPACEL) && !isSPACE_LC_utf8(p)))
+ {
+ match = TRUE;
+ }
+ }
+
+ return (flags & ANYOF_INVERT) ? !match : match;
+}
+STATIC U8 *
+reghop(U8 *s, I32 off)
+{
+ dTHR;
+ if (off >= 0) {
+ while (off-- && s < (U8*)PL_regeol)
+ s += UTF8SKIP(s);
+ }
+ else {
+ while (off++) {
+ if (s > (U8*)PL_bostr) {
+ s--;
+ if (*s & 0x80) {
+ while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80)
+ s--;
+ } /* XXX could check well-formedness here */
+ }
+ }
+ }
+ return s;
+}
+STATIC U8 *
+reghopmaybe(U8* s, I32 off)
+{
+ dTHR;
+ if (off >= 0) {
+ while (off-- && s < (U8*)PL_regeol)
+ s += UTF8SKIP(s);
+ if (off >= 0)
+ return 0;
+ }
+ else {
+ while (off++) {
+ if (s > (U8*)PL_bostr) {
+ s--;
+ if (*s & 0x80) {
+ while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80)
+ s--;
+ } /* XXX could check well-formedness here */
+ }
+ else
+ break;
+ }
+ if (off <= 0)
+ return 0;
+ }
+ return s;
+}